revlog: use an LRU cache for delta chain bases
Profiling using statprof revealed a hotspot during changegroup
application calculating delta chain bases on generaldelta repos.
Essentially, revlog._addrevision() was performing a lot of redundant
work tracing the delta chain as part of determining when the chain
distance was acceptable. This was most pronounced when adding
revisions to manifests, which can have delta chains thousands of
revisions long.
There was a delta chain base cache on revlogs before, but it only
captured a single revision. This was acceptable before generaldelta,
when _addrevision would build deltas from the previous revision and
thus we'd pretty much guarantee a cache hit when resolving the delta
chain base on a subsequent _addrevision call. However, it isn't
suitable for generaldelta because parent revisions aren't necessarily
the last processed revision.
This patch converts the delta chain base cache to an LRU dict cache.
The cache can hold multiple entries, so generaldelta repos have a
higher chance of getting a cache hit.
The impact of this change when processing changegroup additions is
significant. On a generaldelta conversion of the "mozilla-unified"
repo (which contains heads of the main Firefox repositories in
chronological order - this means there are lots of transitions between
heads in revlog order), this change has the following impact when
performing an `hg unbundle` of an uncompressed bundle of the repo:
before: 5:42 CPU time
after: 4:34 CPU time
Most of this time is saved when applying the changelog and manifest
revlogs:
before: 2:30 CPU time
after: 1:17 CPU time
That nearly a 50% reduction in CPU time applying changesets and
manifests!
Applying a gzipped bundle of the same repo (effectively simulating a
`hg clone` over HTTP) showed a similar speedup:
before: 5:53 CPU time
after: 4:46 CPU time
Wall time improvements were basically the same as CPU time.
I didn't measure explicitly, but it feels like most of the time
is saved when processing manifests. This makes sense, as large
manifests tend to have very long delta chains and thus benefit the
most from this cache.
So, this change effectively makes changegroup application (which is
used by `hg unbundle`, `hg clone`, `hg pull`, `hg unshelve`, and
various other commands) significantly faster when delta chains are
long (which can happen on repos with large numbers of files and thus
large manifests).
In theory, this change can result in more memory utilization. However,
we're caching a dict of ints. At most we have 200 ints + Python object
overhead per revlog. And, the cache is really only populated when
performing read-heavy operations, such as adding changegroups or
scanning an individual revlog. For memory bloat to be an issue, we'd
need to scan/read several revisions from several revlogs all while
having active references to several revlogs. I don't think there are
many operations that do this, so I don't think memory bloat from the
cache will be an issue.
# zeroconf.py - zeroconf support for Mercurial
#
# Copyright 2005-2007 Matt Mackall <mpm@selenic.com>
#
# This software may be used and distributed according to the terms of the
# GNU General Public License version 2 or any later version.
'''discover and advertise repositories on the local network
Zeroconf-enabled repositories will be announced in a network without
the need to configure a server or a service. They can be discovered
without knowing their actual IP address.
To allow other people to discover your repository using run
:hg:`serve` in your repository::
$ cd test
$ hg serve
You can discover Zeroconf-enabled repositories by running
:hg:`paths`::
$ hg paths
zc-test = http://example.com:8000/test
'''
from __future__ import absolute_import
import os
import socket
import time
from . import Zeroconf
from mercurial import (
dispatch,
encoding,
extensions,
hg,
ui as uimod,
)
from mercurial.hgweb import (
server as servermod
)
# Note for extension authors: ONLY specify testedwith = 'internal' for
# extensions which SHIP WITH MERCURIAL. Non-mainline extensions should
# be specifying the version(s) of Mercurial they are tested with, or
# leave the attribute unspecified.
testedwith = 'internal'
# publish
server = None
localip = None
def getip():
# finds external-facing interface without sending any packets (Linux)
try:
s = socket.socket(socket.AF_INET, socket.SOCK_DGRAM)
s.connect(('1.0.0.1', 0))
ip = s.getsockname()[0]
return ip
except socket.error:
pass
# Generic method, sometimes gives useless results
try:
dumbip = socket.gethostbyaddr(socket.gethostname())[2][0]
if not dumbip.startswith('127.') and ':' not in dumbip:
return dumbip
except (socket.gaierror, socket.herror):
dumbip = '127.0.0.1'
# works elsewhere, but actually sends a packet
try:
s = socket.socket(socket.AF_INET, socket.SOCK_DGRAM)
s.connect(('1.0.0.1', 1))
ip = s.getsockname()[0]
return ip
except socket.error:
pass
return dumbip
def publish(name, desc, path, port):
global server, localip
if not server:
ip = getip()
if ip.startswith('127.'):
# if we have no internet connection, this can happen.
return
localip = socket.inet_aton(ip)
server = Zeroconf.Zeroconf(ip)
hostname = socket.gethostname().split('.')[0]
host = hostname + ".local"
name = "%s-%s" % (hostname, name)
# advertise to browsers
svc = Zeroconf.ServiceInfo('_http._tcp.local.',
name + '._http._tcp.local.',
server = host,
port = port,
properties = {'description': desc,
'path': "/" + path},
address = localip, weight = 0, priority = 0)
server.registerService(svc)
# advertise to Mercurial clients
svc = Zeroconf.ServiceInfo('_hg._tcp.local.',
name + '._hg._tcp.local.',
server = host,
port = port,
properties = {'description': desc,
'path': "/" + path},
address = localip, weight = 0, priority = 0)
server.registerService(svc)
def zc_create_server(create_server, ui, app):
httpd = create_server(ui, app)
port = httpd.port
try:
repos = app.repos
except AttributeError:
# single repo
with app._obtainrepo() as repo:
name = app.reponame or os.path.basename(repo.root)
path = repo.ui.config("web", "prefix", "").strip('/')
desc = repo.ui.config("web", "description", name)
publish(name, desc, path, port)
else:
# webdir
prefix = app.ui.config("web", "prefix", "").strip('/') + '/'
for repo, path in repos:
u = app.ui.copy()
u.readconfig(os.path.join(path, '.hg', 'hgrc'))
name = os.path.basename(repo)
path = (prefix + repo).strip('/')
desc = u.config('web', 'description', name)
publish(name, desc, path, port)
return httpd
# listen
class listener(object):
def __init__(self):
self.found = {}
def removeService(self, server, type, name):
if repr(name) in self.found:
del self.found[repr(name)]
def addService(self, server, type, name):
self.found[repr(name)] = server.getServiceInfo(type, name)
def getzcpaths():
ip = getip()
if ip.startswith('127.'):
return
server = Zeroconf.Zeroconf(ip)
l = listener()
Zeroconf.ServiceBrowser(server, "_hg._tcp.local.", l)
time.sleep(1)
server.close()
for value in l.found.values():
name = value.name[:value.name.index('.')]
url = "http://%s:%s%s" % (socket.inet_ntoa(value.address), value.port,
value.properties.get("path", "/"))
yield "zc-" + name, url
def config(orig, self, section, key, default=None, untrusted=False):
if section == "paths" and key.startswith("zc-"):
for name, path in getzcpaths():
if name == key:
return path
return orig(self, section, key, default, untrusted)
def configitems(orig, self, section, *args, **kwargs):
repos = orig(self, section, *args, **kwargs)
if section == "paths":
repos += getzcpaths()
return repos
def configsuboptions(orig, self, section, name, *args, **kwargs):
opt, sub = orig(self, section, name, *args, **kwargs)
if section == "paths" and name.startswith("zc-"):
# We have to find the URL in the zeroconf paths. We can't cons up any
# suboptions, so we use any that we found in the original config.
for zcname, zcurl in getzcpaths():
if zcname == name:
return zcurl, sub
return opt, sub
def defaultdest(orig, source):
for name, path in getzcpaths():
if path == source:
return name.encode(encoding.encoding)
return orig(source)
def cleanupafterdispatch(orig, ui, options, cmd, cmdfunc):
try:
return orig(ui, options, cmd, cmdfunc)
finally:
# we need to call close() on the server to notify() the various
# threading Conditions and allow the background threads to exit
global server
if server:
server.close()
extensions.wrapfunction(dispatch, '_runcommand', cleanupafterdispatch)
extensions.wrapfunction(uimod.ui, 'config', config)
extensions.wrapfunction(uimod.ui, 'configitems', configitems)
extensions.wrapfunction(uimod.ui, 'configsuboptions', configsuboptions)
extensions.wrapfunction(hg, 'defaultdest', defaultdest)
extensions.wrapfunction(servermod, 'create_server', zc_create_server)